Reversed Plane Controls, SAS Nosediving

[Landwalker]

I've never been particularly adept at designing planes in KSP, but I've never had this much trouble before.

In my current career game, I have a contract for a low-altitude pressure scan. No problem. I head into the SPH to put together a simple little unmanned drone with a barometer to fly out and take the measurement.

The problems are two-fold:

1. As it's currently designed, the pitch controls are upside-down. W causes it to pitch up, and S causes it to pitch down. The roll controls (Q and E) work normally. If I move things around and cause the Center of Mass to be behind the Center of Lift, the pitch controls operate normally... but the roll controls flip instead.
   
2. Regardless of the wing positioning or the WS/QE control inversions, turning on SAS (regardless of whether or not the HECS probe core has torque enabled or disabled) causes the plane to immediately nosedive.

Neither of these are situations I've ever run across before. Can any of the more competent folk help me figure out what the problem is and how to solve it?

Edited July 6, 2015 by Landwalker

[Red Iron Crown]

The SAS logic thinks your control surfaces are ahead of the CoM, that's why the pitch controls are reversed (and why SAS flips out when turned on). I'd turn off pitch on the ailerons and make the tail surfaces control pitch.

[Landwalker]

The SAS logic thinks your control surfaces are ahead of the CoM, that's why the pitch controls are reversed (and why SAS flips out when turned on). I'd turn off pitch on the ailerons and make the tail surfaces control pitch.

Interesting... I wonder if SAS logic changed between 1.0.2 and 1.0.4 (since I've never seen that behavior before).

Regardless, it makes sense. I jumped into a sandbox and put together an identical plane, but using Delta-Deluxe Winglets on the tails instead of the fixed ones in my original screenshot, and did as suggestedâ€â€disabled pitch on the "main-wing" Elevons (and also disabled roll on the Delta-Deluxe Winglets on the tail, just for the hell of it), and the new plane flies as much like a charm as anything I've built in the past, so that seems to have been the ticket. Thank you kindly for the help.

[Merandix]

If you push the stick forward, the nose is intended to pitch down. This is usually accomplished by placing control surfaces controlling pitch at the extremities of the plane... so at the tail (tailplane) or at the nose (canards) (forgive the (possibly) inaccurate terminology, I'm just a guy who learnt this stuff through KSP, not an expert).

You don't have canards, so typically for a tailplane this happens:

If you push the stick forward, and you have a tailplane with control surfaces, these will change the shape of your tailplane so your tail basically gets more lift, and gets pushed up. If your tail goes up, your nose goes down. Because your main wing isn't doing any deflecting, you will pitch down.

However, if you have ONLY control surfaces on your main wing, and those control surfaces are BEHIND the CoM, your SAS will figure, hey... these control surfaces MUST be a tailplane! So, you press forward (push stick forward, intending to push the nose down), your control surfaces will move down to increase lift, since this must be a tailplane according to the SAS. So now, your main wing is generating more lift, but your tailplane is neutral here... so you will pitch up instead. of the expected pitch down.

This also means that your SAS thinks you should be pitching up when you are actually pitching down. But because the nose goes down, your SAS goes into panic mode, and pulls up harder... which fails because the pitch controls are inverted and only makes it worse.

Again, for effectiveness, the pitch controls are on the extremities of the plane, but work opposite. Because your plane is a see-saw one end moves up, the other end moves down. This is what the SAS assumes. And I believe it uses the CoL as a pivot point.

Why the roll reverses when you put the CoL in front of the CoM, I don't know.

On this design it stands out that there's no yaw authority in it's entirety. Both your tailplane and vertical stabilisers seem to be AV-T1 winglets, which have no control surfaces.

So to fix this, remove pitch from the control surfaces on your main wing, and add control surfaces on your tailplane (disable roll and yaw here). I'd also recommend putting a control surface on your vertical stabiliser for yaw control (disable pitch and roll here).

[Landwalker]

[bunch of stuff that makes sense]

You observe correctlyâ€â€the tail of the craft is entirely fixed (and is entirely AV-T1), since at the time of design I hadn't unlocked any controllable surfaces that were suitable for the tail. My attitude towards the tail was mostly "It's here to move the CoL back a little bit and get some stabilizing drag going," which was (clearly) not a very effective or useful attitude. I've since unlocked the Delta-Deluxe Winglet part, which also happens to be what I used in my successful sandbox test. That part will make up the preponderance of my tail-based assemblies for the foreseeable future, I think.

It sounds like I may have just gotten lucky in the past and had a plane that was so nose-heavy that there wasn't any doubt about where the SAS thought the wings were in relation to the CoM, but that this particular assignment resulted in me stumbling upon the more normal arrangement.

Thanks for the recommendations regarding where and how to limit control authority. In the past I've just sort of slapped on control surfaces and hoped it all worked, so I'll give these a shot going forward and hopefully have better luck / less guess-and-check and more just-do-it-right.